Abstract
The World Health Organisation reports distracted driving actions as the main cause of road traffic accidents. Current studies to detect distraction postures focus on analysing spatial features of images using Convolutional Neural Networks (CNN). However, approaches addressing both spectral and spatial features of images for driving distraction are scarce. Our hypothesis is that deep learning approaches can further be exploited to consider spatial and spectral features, so that the spatial features capture the spatial information within the image and the spectral features capture the spectral correlations among the image channels. This paper introduces a novel driver distraction posture detection method using CNNs and stacked Bidirectional Long Short Term Memory (BiLSTM) Networks to capture the spectral-spatio features of the images. The proposed methodology consists of two stages: first, we automatically learn the spatial posture features using pre-trained CNNs. Subsequently, we utilise BiLSTMs architecture to extract the spectral features amongst the stacked feature maps from the pre-trained CNNs. Our proposed approach is evaluated on the American University in Cairo (AUC) Distracted Driver Dataset, the most comprehensive and detailed dataset on driver distraction postures to date. Results show that our approach beats state-of-the-art CNN models with an average classification accuracy of 92.7%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.